|
| Selection of Near Infrared Spectral Wavelength Variables Based on Improved Whale Optimization Algorithm and Its Application |
| |
| View Full Text Download reader |
| DOI： |
| KeyWord:near-infrared spectroscopy wavelength selection improved whale optimization algorithm transfer function greedy thinking |
| Author | Institution |
| WANG Zhong-yu，GAO Mei-feng |
Key Laboratory of Advanced Process Control for Light Industry（Ministry of Education），School of Internet of Things Engineering，Jiangnan University，Wuxi ，China |
|
| Hits: 1430 |
| Download times: 1100 |
| Abstract: |
| Based on the whale optimization algorithm（WOA） of swarm intelligence，an improved whale optimization algorithm（iWOA） for the selection of near-infrared spectral wavelengths was proposed. Firstly，a chaotic strategy was introduced to initialize the population to avoid the algorithm from falling into local optimization prematurely. Secondly，a nonlinear time-varying Sigmoid transfer function and greedy algorithm were introduced to improve the algorithm's optimization ability and make the model obtain better prediction accuracy. In order to verify the effectiveness of the algorithm，the near-infrared spectral data of four indicators for corn fat，protein，starch and water were used for PLS modeling and analysis，and compared with other algorithms. The results showed that the iWOA algorithm could effectively filter out the wavelength variable in the shortest possible time，reduce the complexity of the model and improve the prediction accuracy of the model. The root mean square errors of prediction（RMSEPs） of the model decreased from 0.077 2，0.122 4，0.334 4 and 0.059 5 to 0.033 2，0.050 7，0.139 2 and 0.004 4，and the prediction accuracy was improved by 57.0%，58.6%，58.3% and 92.6%，respectively，compared with those of the full spectrum. The numbers of wavelengths selected by the algorithm were 84，69，87 and 66，respectively. |
| Close |